Optimal real-time scheduling for observations of transients across global telescope networks

遍布世界的机器人望远镜网络为时间域天文学制度开辟了新的机会。这类网络的部署伴随着新一代强大调查的推出，这些调查提供“临时工厂”，要求迅速和广泛的后续监测。如何有效利用这些资源，如何协调观测？虽然到目前为止，望远镜的调度主要是从服务提供商的角度进行讨论，但我们需要从希望使用来自不同提供商的各种资源来实现科学目标的用户社区的角度来看待这一问题。在选择由特定仪器在特定时间观察的特定目标时，需要考虑到：1)科学目标，2)关于所有潜在相关目标的已获得数据，3)仪器的技术规格，4)目标的可观察性。特别是，目标优先顺序的变化速度与目标本身的变化速度一样快，因此需要准实时地分析数据，以便不遗漏关键特征。因此，需要立即将大量数据集中在一起并进行处理，而持续监控需要不间断的可操作性。在较大的用户群体内，目标职能和由此产生的监测战略还需要考虑数据的所有权以及个人和集体利益。圣安德鲁斯大学率先利用LCOGT/SUPAScope和MindStep网络实施了自动目标选择战略，以跟踪正在进行的重力微透镜事件。学生将在天文学和计算机科学的交叉点面临数据处理、建模和管理方面的技术挑战，以满足实现快速吞吐量的要求。这类问题的解决方案可能会被移植到其他领域的应用程序中。

World-spanning networks of robotic telescopes have opened up new opportunities for the regime of time-domain astronomy. The deployment of such networks comes along with the roll-out of a new generation of powerful surveys that provide "transient factories", calling for prompt and extensive follow-up monitoring. How can such resources be used efficiently and how can observations be coordinated? While the scheduling of telescopes so far has mostly been discussed from the perspective of service providers, we need to look at this from the perspective of the community of users who want to achieve a scientific objective using a diversity of resources coming from different providers. The selection of a specific target to be observed at a specific time by a specific instrument needs to take into account: 1) the science goals, 2) the acquired data on all potentially relevant targets, 3) the technical specifications of the instrument, 4) the observability of the targets. In particular, target priorities will change as fast as the targets vary themselves and data therefore need to be analysed in quasi-real time in order not to miss critical features. Consequently, a large amount of data need to be pulled together and processed immediately, while continuous monitoring requires uninterrupted operability. Within larger user communities, objective functions and resulting monitoring strategies also need to consider the ownership of data and both individual and collective benefits. The University of St Andrews has pioneered the implementation of automated target selection strategies for the follow-up of ongoing gravitational microlensing events with the LCOGT/SUPAscope and MiNDSTEp networks. Students will face a technology challenge on data processing, modelling, and management at the intersection of astronomy and computer science, matching the requirement to achieve a fast throughput. Solutions to this sort of problem will likely to be transferable to serve applications in other areas.